# GDS Agent
Neither LLMs nor any existing toolings (MCP Servers) are capable of complex reasoning on graphs at the moment.
This MCP Server includes toolings from Neo4j Graph Data Science (GDS) library, which allows you to run all common graph algorithms.
Once the server is running, you are able to **ask any graph questions about your Neo4j graph** and get answers. LLMs equipped with GDS agent can decide and accurately execute the appropriate parameterised graph algorithms over the graph you have in your Neo4j database.
An example where an LLM with GDS Agent is able to pick shortest path and Yen's algorithm to answer my question about travel plan:
# Usage guide
If you have `uvx` [installed](https://docs.astral.sh/uv/getting-started/installation/), add the following config to your `claude_desktop_config.json`
```
{
"mcpServers": {
"neo4j-gds": {
"command": "/opt/homebrew/bin/uvx",
"args": [ "gds-agent" ],
"env": {
"NEO4J_URI": "bolt://localhost:7687",
"NEO4J_USERNAME": "neo4j",
"NEO4J_PASSWORD": ""
}
}
}
}
```
Replace command with your `uvx` location. Find out by running `which uvx` in the command line.
Replace `NEOJ_URI`, `NEO4J_USERNAME`, `NEO4J_PASSWORD` with your database login details. You can also optionally specify `NEO4J_DATABASE`.
# Full documentation
For complete documentation and development guidelines, please refer to: https://github.com/neo4j-contrib/gds-agent.
Raw data
{
"_id": null,
"home_page": null,
"name": "gds-agent",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.11",
"maintainer_email": null,
"keywords": "graph, graph algorithms, graph data science, mcp, neo4j",
"author": null,
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/09/89/877d77e383fa0edabdf86000b12f75e58a62f0d9e3b008c72228cb2a07e3/gds_agent-0.1.0.tar.gz",
"platform": null,
"description": "# GDS Agent\n\nNeither LLMs nor any existing toolings (MCP Servers) are capable of complex reasoning on graphs at the moment.\n\nThis MCP Server includes toolings from Neo4j Graph Data Science (GDS) library, which allows you to run all common graph algorithms.\n\nOnce the server is running, you are able to **ask any graph questions about your Neo4j graph** and get answers. LLMs equipped with GDS agent can decide and accurately execute the appropriate parameterised graph algorithms over the graph you have in your Neo4j database.\n\nAn example where an LLM with GDS Agent is able to pick shortest path and Yen's algorithm to answer my question about travel plan:\n\n\n\n# Usage guide\nIf you have `uvx` [installed](https://docs.astral.sh/uv/getting-started/installation/), add the following config to your `claude_desktop_config.json`\n```\n{\n \"mcpServers\": {\n \"neo4j-gds\": {\n \"command\": \"/opt/homebrew/bin/uvx\",\n \"args\": [ \"gds-agent\" ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USERNAME\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"\"\n }\n }\n }\n}\n```\nReplace command with your `uvx` location. Find out by running `which uvx` in the command line.\nReplace `NEOJ_URI`, `NEO4J_USERNAME`, `NEO4J_PASSWORD` with your database login details. You can also optionally specify `NEO4J_DATABASE`.\n\n# Full documentation\nFor complete documentation and development guidelines, please refer to: https://github.com/neo4j-contrib/gds-agent.",
"bugtrack_url": null,
"license": "MIT",
"summary": "Neo4j GDS MCP Server",
"version": "0.1.0",
"project_urls": {
"Homepage": "https://github.com/neo4j-contrib/gds-agent"
},
"split_keywords": [
"graph",
" graph algorithms",
" graph data science",
" mcp",
" neo4j"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "91903e6c8db6645980e4a4d2160f5eae1ba6d230bab03c3754174cae0e30586c",
"md5": "418612f9e55de049f12c0f573e835a37",
"sha256": "80d9422e6a4a794cea75442dc200085b8cfa553ad17442339d7489eebda2e679"
},
"downloads": -1,
"filename": "gds_agent-0.1.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "418612f9e55de049f12c0f573e835a37",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.11",
"size": 36450,
"upload_time": "2025-07-14T21:25:07",
"upload_time_iso_8601": "2025-07-14T21:25:07.120019Z",
"url": "https://files.pythonhosted.org/packages/91/90/3e6c8db6645980e4a4d2160f5eae1ba6d230bab03c3754174cae0e30586c/gds_agent-0.1.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "0989877d77e383fa0edabdf86000b12f75e58a62f0d9e3b008c72228cb2a07e3",
"md5": "906a5af31e8c9fea950746cb03b6299a",
"sha256": "49c1dc6b0f48ab55fda43adc402057ef5b2c49bc6fa87037c3e9f9c49de3b926"
},
"downloads": -1,
"filename": "gds_agent-0.1.0.tar.gz",
"has_sig": false,
"md5_digest": "906a5af31e8c9fea950746cb03b6299a",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.11",
"size": 87466,
"upload_time": "2025-07-14T21:25:08",
"upload_time_iso_8601": "2025-07-14T21:25:08.639587Z",
"url": "https://files.pythonhosted.org/packages/09/89/877d77e383fa0edabdf86000b12f75e58a62f0d9e3b008c72228cb2a07e3/gds_agent-0.1.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-07-14 21:25:08",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "neo4j-contrib",
"github_project": "gds-agent",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [
{
"name": "mcp",
"specs": [
[
"==",
"1.8.0"
]
]
},
{
"name": "graphdatascience",
"specs": [
[
"==",
"1.14"
]
]
},
{
"name": "neo4j",
"specs": [
[
"==",
"5.28.1"
]
]
}
],
"lcname": "gds-agent"
}
JSON
